Cross-sectional and High-Temperature Structure-Property Relationships in Nanocrystalline Thin Films

(1)

Dipl.-Ing. Michael Meindlhumer, BSc

Cross-sectional and High-Temperature Structure-Property Relationships in

Nanocrystalline Thin Films

Doctoral Thesis

(2)

(3)

(4)

(5)

I wish to express my deepest thanks and gratitude to all the people who have helped me forming this thesis in the past 4 years, either professionally or personally.

First and foremost I want to thank my advisors, Prof. Jozef Keckes and Prof.

Christian Mitterer. Their continuous support and their constructive feedback on numerous scientific topics are a great source of improvement for my work, as well as for myself. Furthermore, I want to thank Assoz. Prof. Rostislav Daniel for the possibility to perform this thesis within the framework of the Christian Doppler Laboratory for the Advanced Synthesis of Novel Multifunctional Coatings. His support and critical review of the scientific work and during writing of the publications are also greatly acknowledged.

A special thank-you is directed towards Jozef Keckes, who employed me during my bachelor and master studies from 2013 to 2016, and who always gave ear to my questions, and was always enthusiastic about my results and queries. His mentoring also encouraged me to pursue my scientific ideas, some which were even incorporated into this thesis.

Significant parts of this work were carried out at the ID13 beamline of the European Synchrotron (ESRF, Grenoble) and the P07 beamline of the PETRA III lightsource at DESY in Hamburg, Germany. I want to thank the responsible beamline sci- entists, Manfred Burghammer and Martin Rosenthal from the ESRF and Andreas Stark from the P07 at DESY for their excellent support during the measurements and the numerous discussions about topics related to materials science and various other fields of science, life or fiction.

A great number of samples for synchrotron beamtimes and micromechanical testing was prepared by Gabriele Felber, Herwig Felber and Daniela Keckesova in our department’s TEM preparation laboratory. Their efforts and skill were essential for the success of this work.

Among my former and present colleagues I would like to thank especially Sabine Bodner, David Gruber, Julius Keckes, Kevin Kutlesa, Michael Reisinger, Juraj Todt Jakub Zalesak, Tobias Ziegelwanger, and all the others, I might have missed here, for the (sometimes) thriving office discussions, partly even related to materials science, their support and generally all the other little distractions.

I want to express my thanks to all my friends here in Leoben, especially Markus, Andrea, Severin and Julia for all the time spent having fun together. Additionally, I very much appreciate the time spent with my co-musicians from my chamber music

(6)

ensemble Música Atmósfera. Thank you for leading me astray from materials science sometimes.

Furthermore, I want to thank my brothers Martin and Andreas, as well as their fiancées, Martina and Anna. Especially, I want to thank my parents Gabriele and Erwin, without their support in every issue, this thesis would not have been written.

Last but most importantly, I want to thank my girlfriend Julia, for her uncondi- tional love, her constant support and all the joy she brings to my life, including also (a little more) chaos.

(7)

financially supported by Christian Doppler Research Association. The financial support by the Austrian Federal Ministry of Science, Research and Economy and the National Foundation for Research, Technology and Development is also gratefully acknowledged.

(8)

(9)

Microstructure, mechanical and functional properties of thin films often exhibit gradients at the nanoscale, which originate either from the nonuniform vapour deposition processes or were introduced in the film via external loads. Independent of their origin, resolving thin films gradients of interest, needs characterisation tools operating with a spatial resolution at the nanoscale. These gradients of microstructure, residual and applied strains in hard protective thin films are the focus of this work, since they are critical to the performance of coated structures, such as coated tools used in the machining industry.

Three model problems within thin film research were approached within this thesis, including (i) the nanoscale microstructure and residual stress gradients emer- ging from the deposition process in a thin film coated onto a cutting edge area, (ii) the nanoscale mechanical response against scratching and (iii) the in situ evaluation of the fracture response of thin films. Moreover, (iv) the thorough investigation of decomposition of thin films during annealing up to 1100°C, and finally (v) a self-assembled hierarchical thin film with a superior combination of mechanical and thermal properties is presented. In detail, the following characterisations on various thin films were performed:

• Cross-sectional X-ray nanodiffraction was applied to access the nanoscale microstructure and stress gradients originating from the physical vapour deposition process in a TiN coating on a WC-Co cutting edge. While gradual and constant residual stress distributions with magnitudes between −1.4 and

−2.4GPa were found at the flank and rake faces, respectively, directly at the cutting edge a pronounced lateral and cross-sectional gradient ranging from 0 to−3GPa was evaluated from the X-ray data.

• 50 nm X-ray nanodiffraction and alectron microscopy were applied to charac- terize the nanoscale stress gradients and microstructural changes across the scratch track area of a brittle-ductile Cr/CrN thin film on a high speed steel substrate. After scratching, the formation of severe nanoscopic gradients of in-plane, out-of-plane and shear stress distributions were revealed in Cr and CrN, ranging from −6 to 1.5 GPa. These stress gradients were accompanied by and correlated to irreversible microstructural changes, such as either inter- granular crack formation and transgranular defects, or crystallite bending and uniaxial gliding in CrN and Cr, respectively.

(10)

• The fracture response of a notched clamped cantilever composed of four alternating Cr and CrN layers on high speed steel was evaluated by in situ cross-sectional X-ray nanodiffraction. The residual stress distributions in the notched Cr layer result in an effective stress intensity factor of

−5.9±0.4MPa m¹^/² and a pronounced stress concentration and a plastic zone around the notch. At a critical stress intensity of 2.8±0.5MPa m¹^/², crack growth occurred up to the adjacent CrN-Cr interface, where the crack was arrested.

• The stress-controlled decomposition routes of three AlCrN thin films have been assessed by the newly developed in situ high-temperature high-energy grazing-incidence-transmission X-ray diffraction method. Whereas the decomposition temperatures of the metastable cubic Al_0.7Cr_0.3Nphase ranged from 698-914°C, the residual stress level of ∼ −4.3GPa was similar for all three investigated thin films.

• A TiAlN thin film composed of 6 hierarchical levels and mimicking biological materials, such as nacre or enamel, was synthesized by chemical vapour deposition, by alternating two variants of chemical precursors. In such a way, an irregular multilayer stack was formed by bottom-up self-assembly, consisting of hard herringbone stacks separated by interlayers of spherical nanograins. It exhibits superior functional properties and represents a milestone in the field of synthesis of protective wear resistant thin films.

(11)

Die Mikrostruktur, die mechanischen und funktionellen Eigenschaften von dünnen Schichten haben oft nanoskalige Gradienten, welche entweder im inhomogenen Gasphasen-Abscheideprozess oder durch externe Belastung entstehen. Unabhängig ihrer Herkunft benötigt das Erfassen dieser Gradienten Untersuchungsmethoden mit einer nanoskopischen räumlichen Auflösung. Die Gradienten von Mikrostruktur, Eigen- und angelegter Spannungen in harten, dünnen Schutzschichten sind der Fokus dieser Arbeit, denn diese sind kritisch für die Leistungsfähigkeit von beschichteten Strukturen, wie z.B. beschichtete Werkzeuge in der spanenden Bearbeitung.

Drei Modellprobleme in der Dünnschichtforschung wurden in dieser Arbeit un- tersucht, einschließlich (i) der nanoskaligen Mikrostruktur- und Eigenspannungs- gradienten, welche im Bereich einer Schneidkante durch den Beschichtungsprozess auftreten, (ii) der nanoskaligen mechanischen Antwort auf den Ritzversuch und (iii) der in situ Auswertung der Bruchantwort von Dünnschichten. Zudem wurde (iv) die Analyse der Zersetzung von Dünnschichten bis 1100°C durchgeführt und ab- schließend (v) wird eine selbstorganisierte, hierarchisch aufgebaute Dünnschicht mit einer überragenden Kombination von mechanischen und thermischen Eigenschaften präsentiert. Folgende Charakterisierungen wurden an Dünnschichten durchgeführt:

• Röntgen-Nanobeugung wurde angewandt um die nanoskopischen Mikrostruktur- und Eigenspannungsgradienten in einer TiN Beschichtung auf einer WC-Co Schneidkante, entstanden durch den Abscheidungsprozess, zu charakterisieren . Während graduelle und konstante Eigenspannungsverteilun- gen mit einer Größe zwischen−1.4und−2.4GPa an der Frei- bzw. Spanfläche gemessen wurden, wurden direkt an der Kante ausgeprägte laterale und Quer- schnittsgradienten in einem Bereich von 0 bis −3GPa aus den Röntgendaten ausgewertet.

• Röntgen-Nanobeugung und Elektronenmikroskopie wurden eingesetzt um die nanoskopischen Eigenspannungs- und Mikrostrukturänderungen in der Umge- bung einer Ritzspur in einer Dünnschicht bestehend aus spröd-duktilem Cr/CrN auf Schnellarbeitsstahl zu charakterisieren. Nach dem Ritztest wurden ausgeprägte Spannungsgradienten in CrN und Cr festgestellt, welche in einem Bereich zwischen −6 und 1.5 GPa liegen. Diese wurden irreversiblen Mikrostrukturänderungen, wie interkristalliner Rissbildung und transkristal- liner Defektansammlung, sowie Kristallitbiegung und gerichtetem Abgleiten in

(12)

CrN bzw. Cr zugeordnet.

• Die Bruchantwort eines gekerbten, beidseitig eingespannten Biegebalkens aus vier abwechselnden Cr- und CrN-Lagen auf Schnellarbeitsstahl wurde mittels in situ Röntgen-Nanobeugung ausgewertet. Die Eigenspannungen in der gekerbten Cr-Schicht ergaben eine effektive Spannungsintensität von

−5.9±0.4MPa m¹^/², begleitet von einer plastischen Zone rund um die Kerbe.

Eine kritische Spannungsintensität von 2.8±0.5MPa m¹^/² führte zu Rissaus- breitung an die anliegenden CrN-Cr Grenzfläche, wo diese gestoppt wurde.

• Die spannungskontrollierten Zersetzungsrouten in drei AlCrN Dünnschichten wurden mittels der neu entwickelten in situ Hochtemperatur-Hochenergie- Durchstrahlungs-Röntgenbeugung festgestellt. Obwohl die Zersetzungstem- peraturen des metastabilen kubischen Al0.7Cr0.3N zwischen 698 und 914°C variieren, ist die Eigenspannung zu Beginn der Zersetzung mit ∼ −4.3GPa für alle drei untersuchten Dünnschichten gleich.

• Eine TiAlN Dünnschicht mit 6 hierarchischen Ebenen, welche Biomaterialien imitiert, wurde mittels chemischer Gasphasenabscheidung durch Abwechslung zweier verschiedener Ausgangsstoffvarianten hergestellt. So wurde ein irregulärer Multilagenaufbau durch Selbstorganisation hergestellt, bestehend aus harten fischgrätenartig aufgebauten Lagen, welche durch Zwischenlagen sphärischer Nanokörner getrennt werden. Diese Beschichtung zeigt überra- gende funktionale Eigenschaften und repräsentiert einen Meilenstein im Bereich der Herstellung harter, verschleißfester Schutzschichten.

(13)

Affidavit III

Acknowledgments V

Abstract IX

Kurzfassung XI

1. Introduction 1

1.1. Deposition-induced gradients of microstructure and residual stress

along the cutting edge . . . 2

1.2. Gradients of microstructure and stresses induced by external loads . 3 1.3. Aim of the thesis . . . 5

2. Hard protective thin films 9 2.1. TiN . . . 10

2.2. Cr/CrN multilayer thin films . . . 10

2.3. AlCrN . . . 11

2.4. AlTiN . . . 12

3. Selected advanced characterization techniques 15 3.1. X-ray diffraction . . . 15

3.1.1. Cross-sectional X-ray nanodiffraction . . . 17

3.1.2. In situ high-energy high-temperature grazing incidence transmission X-ray diffraction . . . 31

3.2. Micromechanical testing . . . 33

3.2.1. Nanoindentation . . . 33

3.2.2. In situ micromechanical cantilever bending experiments . . . 36

4. Conclusions and Outlook 43 5. List of appended publications 57 5.1. Papers in scientific journals . . . 57 A. Nanoscale residual stress and microstructure gradients across the cut-

ting edge area of a TiN coating on WC-Co A–1

(14)

B. Nanoscale stress distributions and microstructural changes at scratch track cross-sections of a deformed brittle-ductile CrN-Cr bilayer B–1 B.1. Introduction . . . B–2 B.2. Experimental . . . B–4 B.2.1. CrN/Cr thin film synthesis . . . B–4 B.2.2. Scratch testing . . . B–4 B.2.3. Sample preparation . . . B–5 B.2.4. CSnanoXRD analysis . . . B–5 B.2.5. Simulation . . . B–7 B.3. Results and Discussion . . . B–8 B.3.1. Cross-sctional scratch track area morphologies . . . B–8 B.3.2. Small-angle X-ray scattering microscopy . . . B–12 B.3.3. 2D FWHM Analysis . . . B–13 B.3.4. Qualitative texture analysis . . . B–15 B.3.5. Stress analysis across scratch track cross-sections . . . B–18 B.3.6. Finite Element Model . . . B–24 B.4. Discussion of brittle-ductile CrN-Cr bilayer deformation . . . B–27 B.5. Conclusions . . . B–29 C. Evolution of stress fields during crack growth and arrest in a brittle-

ductile CrN-Cr clamped-cantilever analysed by X-ray nanodiffraction

and modelling C–1

C.1. Introduction . . . C–2 C.2. Experiment and methods . . . C–4 C.2.1. Thin film synthesis . . . C–4 C.2.2. FIB-preparation and investigation of the cantilever . . . C–5 C.2.3. CSnanoXRD experiment . . . C–5 C.2.4. CSnanoXRD data analysis . . . C–7 C.2.5. 2D-Simulation . . . C–9 C.3. Results . . . C–10

C.3.1. Ex situ thin film analysis . . . C–10 C.3.2. In situ experiment . . . C–14 C.3.3. Simulated stress results . . . C–27 C.4. Discussion . . . C–29 C.4.1. In situ synchrotron setup and measurement conditions . . . . C–29 C.4.2. Stress state around the crack tip in as-fabricated state . . . . C–30 C.4.3. Loading, crack growth, crack tip blunting and crack closing . C–31

(15)

C.5. Conclusions . . . C–32 D. Stress-controlled decomposition routes in cubic AlCrN films assessed

by in-situ high-temperature high-energy grazing incidence transmission

X-ray diffraction D–1

D.1. Introduction . . . D–2 D.2. Results . . . D–4 D.2.1. In-Situ Phase Analysis . . . D–4 D.2.2. In-Situ Qualitative Texture Analysis . . . D–6 D.2.3. Full Width at Half Maximum Analysis . . . D–10 D.2.4. Unstrained Lattice Parameter Analysis . . . D–11 D.2.5. Experimental Thermal Expansion Coeﬃcients and Thermal

Strains . . . D–13 D.2.6. In-situ Residual Strain and Stress Evolution . . . D–14 D.2.7. Experimental In-Plane Intrinsic Strains . . . D–16 D.2.8. Complementary Analyses . . . D–16 D.3. Discussion . . . D–18 D.4. Conclusions . . . D–23 E. Biomimetic hard and tough nanoceramic Ti–Al–N film with self-assembled

six-level hierarchy E–1

E.1. Introduction . . . E–2 E.2. Results . . . E–3 E.2.1. Self-assembly of hierarchical microstructure . . . E–3 E.2.2. Micro- and nanomechanics . . . E–6 E.2.3. High-temperature stability . . . E–11 E.3. Discussion and Conclusions . . . E–12

List of co-authored publications I

(16)

(17)

1

Introduction

Nowadays, thin films are essential in many state-of-the art technologies, such as functional layers in electronic devices or wear-protective coatings for high-speed cutting tools. Especially, in the case of hard protective thin films deposited on cutting tools, research and development have been stimulated by demands from machining industry over the recent years [1–5]. The films protecting cutting tools have to withstand severe conditions during operation, which are mainly high mechanical loads such as friction-induced stress [6, 7] and temperatures up to 1000°C [8] directly at the cutting edge. These films must exhibit outstanding mechanical properties, which allow them resisting the thermal, abrasive and mechanical loading conditions during operation, as summarized above. Required film properties include mainly high hardness, Young’s modulus, fracture toughness, as well as wear and oxidation resistance [9, 10]. Additionally, growth defects originating from the deposition process [11], such as droplets, can act as stress concentrators during loading, which can initiate premature failure of the thin film [12].

Multiple gas phase techniques have been developed to deposit thin ﬁlms, which can be roughly separated into two categories, physical (PVD) and chemical vapour deposition (CVD).

Typical PVD techniques are cathodic arc evaporation (CAE) and magnetron sput- ter deposition (MSD). Both CAE and MSD are plasma-assisted deposition techniques operating far from the thermodynamic equilibrium, which allows deposition of mixtures of immiscible elements [13–15].

In detail, CAE utilizes arc discharges to vaporize material from the surface of an electrode (the cathode) [15, 16]. Using CAE has some advantages, like high deposition rates and generally a good adhesion of the ﬁlm to the substrate [15].

However, the main drawback is the frequent formation of macroparticles, known as droplets, which are generally detrimental to the ﬁlm’s quality and properties [11, 12].

The main advantages of MSD are that no macroparticles are formed during deposition and that the surface and interface roughnesses are signiﬁcantly lower compared

(18)

to CAE and CVD [13, 14]. However, deposition rates are rather low compared to CAE or CVD.

On the other hand, during CVD [17], gaseous precursors form a solid ﬁlm on the substrate surface via chemical gas phase reactions, condensation from the vapour phase and further surface reactions. The advantages of CVD are the high deposition rates and that it is possible to form ﬁlms even on geometrically complex structures, however the main drawback is that high deposition temperatures of ∼700-1100°C are necessary [17, 18].

Thin films synthesized by PVD or CVD normally exhibit complex cross-sectional and lateral nanoscale gradients of microstructure and physical properties, as a consequence of the non-equilibrium preparation process [19–21]. These gradients are a result of the gradual development of crystalline and amorphous phases, the crystallite’s size and shape, crystallographic texture and strains of 1^st, 2^nd and 3^rd order developed during deposition of the thin film. In order to guide the functional properties of thin films, it is necessary to understand the specific local structure-property relationships induced by the non-equilibrium deposition processes. All of the properties necessary for hard protective thin films stated above, are decisively influenced by the gradients of microstructure and residual stress predefined by the deposition process, which in turn can be altered deliberately by tuning particular process parameters [20–22].

In the following, the state-of-the art in the analysis of (i) deposition-induced gradients of microstructure and residual stress and (ii) the nanoscale stress gradients originating in the thin ﬁlms from external loads are brieﬂy addressed and (iii) the aim of this thesis is outlined.

1.1. Deposition-induced gradients of microstructure and residual stress along the cutting edge

As already mentioned above, especially the cutting edge area, which is in direct contact with the workpiece, undergoes heavy mechanical and thermal loads during operation [6, 7, 23]. Consequently, design strategies of thin ﬁlms controlling chemical composition, microstructure, residual stress, thermal stability and mechanical properties should be focused exactly on this area. However, up to now, optimisa- tion of the deposition process in terms of the loop of design, synthesis and property characterisation is almost exclusively performed on ﬂat surfaces. This systematic inconsistency completely neglects the nanoscale gradients of chemical composition,

(19)

microstructure, thickness and morphology evaluated by previous studies for PVD thin ﬁlms at the cutting edges [24–27], which differ significantly from the ones observed on flat surfaces. In detail, a thickness increase of up to 26% and a decrease of the Al/Ti ratio up to 14% was evaluated for a TiAlCrYN coating on sharp WC-Co edges with varying opening angles by Macak et al. [25]. On a milling tool coated with TiAlN, stress relaxation from −5.5 to−2.5GPa was observed from the centre to the edge by laboratory X-ray diffraction (XRD) [28].

Generally, these findings were related to the evolution of a spatially non- homogeneous process plasma around the substrate edges and a locally enhanced deposition temperature [24–26, 29]. Given the observed change in microstructure, chemical composition and morphology revealed by Refs. [24–26, 29], tremendous nanoscale lateral and cross-sectional changes of microstructure and residual stress are to be expected in coatings in cutting edge areas, after deposition. The nanoscale evaluation of these gradients, whose origin lies in the deposition process itself, could lead to further coating optimization, detect the origins of premature film failure and eventually prolong the lifetime of coated cutting tools.

1.2. Gradients of microstructure and stresses induced by external loads

Alongside nanoscale elemental, microstructural and mechanical/stress gradients originating from the deposition process, additional microstructural and stress gradients are introduced in a coating during service, which are mostly related to either thermal, mechanical or abrasive loads applied to the coating. Generally, abrasive loads can be simulated by tribological tests, such as the ball-on-disk test using a sphere of an abrasive material, which slides over the coating surface [30], or the scratch test, where an indenter tip is moved uniaxially over the coating’s surface [31, 32]. The scratch test itself was initially developed to qualitatively assess the adhesion of hard ceramic thin films on various types of substrates. Subsequently it was revealed, that by changing a variety of sample (intrinsic) and method (extrinsic) related parameters, the testing mode can be altered in such way, that not the adhesion of a film is decisive, but where deformation/cohesive strength of the film itself is crucial [33–

35]. The latter is a model case for the single-asperity contact, e.g. for a metal chip sliding over the flank face of a coated cutting tool. However, the state-of-the art for assessing structural changes in thin films, induced by scratch testing, is a combination of acoustic emission spectroscopy, analysis of the load-indentation depth curve

(20)

over the scratched distance and optical and electron microscopy of the scratched surface, to identify critical loads at specific damage levels to the film [36]. Up to now, strains/stresses induced by scratch testing have been primarily assessed by finite element modelling [37–39]. However, currently, there is only limited understanding about the microstructural changes and deformation-induced stresses in thin films after application of abrasive loads.

Additionally, externally applied loads during operation in the vicinity of pre- existing defects in hard protective thin films can induce fracture of the coating and subsequently lead to tool failure. Consequently, a sufficiently high fracture toughness is necessary to suppress overcritical brittle fracture, since the fracture toughness itself is a value for the resistance of a material against critical crack growth [40, 41]. Since nanoceramic protective thin films are inherently brittle, they naturally lack fracture toughness, despite their beneficial high hardness and Young’s modulus. Additionally, fracture in thin films progresses along columnar grain boundaries of low cohesive energy, which results in even further decreased fracture resistance [42]. Several approaches have been reported in literature to increase the fracture toughness of thin films, such as multi-layered thin films with constituents exhib- iting pronounced spatial heterogeneity [42–44], alloying with B and C, as well as microstructural design of CVD TiN [45–47], tilting of columnar grain boundaries [2, 48, 49], formation of superlattice thin films [50, 51] and by enhancing plasticity via vacancy formation on the non-metal sublattice [52]. Alternating stiff and com- pliant sublayers is an effective measure to increase the fracture toughness of thin films without major influences on stiffness and hardness, which has its role-models in natural materials, such as nacre, bone or enamel [53–56].

In order to assess the fracture toughness of thin ﬁlms, bending of focused ion beam (FIB) fabricated single-clamped micro-cantilevers has evolved into the state- of-the art methodology for testing of hard protective thin ﬁlms [2, 42, 43, 45–49, 57].

However, in recent years, also the testing of double-clamped cantilevers, manufactured by FIB has drawn significant attention [58–60]. In single-edged cantilevers the mechanical equilibrium condition has to be fulfilled at the freestanding end, which relieves nearly all residual stress contributions from the film. On the contrary, in the double-clamped cantilever geometry, the residual strain from deposition is nearly fully preserved, since the cantilever has no freestanding end. This makes the double- clamped cantilever geometry a suitable and interesting candidate to investigate the influence of residual stresses on the fracture response of thin films.

In summary, despite signiﬁcant progress in the development of thin ﬁlms over the recent years, it is still necessary to evaluate and understand the formation of

(21)

local mechanical and microstructural gradients of thin ﬁlms at the nanoscale, either introduced by the deposition process or by externally applied loads.

1.3. Aim of the thesis

The aim of the present thesis is focussed on examining and evaluating structure- property relationships of hard protective thin films at the nanoscale and at elevated temperatures. In detail, the five enclosed publications deal with the characterization of nanoscale stress and microstructure gradients in thin films (i) deposited onto industrially manufactured cutting-edge WC-Co substrates, (ii) after scratch testing, which is a model of the single-asperity contact, and (iii) during in situ micromechanical loading. Moreover, (iv) a new methodology evaluating the crucial influence of residual stress on the decomposition of metastable AlCrN thin films and a (v) novel six-level hierarchical thin film with superior fracture resistance are presented.

In detail, within this thesis, several aspects of the structure-property relationships of thin ﬁlms have been investigated, such as:

1. The nanoscale microstructural and residual stress gradients of hard protective thin ﬁlms are of high scientiﬁc and industrial interest. In Paper A, cross- sectional X-ray nanodiffraction with a beam size of 35 × 25 nm² was used to retrieve structural and mechanical gradients in the cutting edge area of a

∼ 2µm thick TiN thin film on WC-Co substrate. Within this work, scanning small-angle X-ray scattering microscopy (SAXSM) was presented and utilized for the first time to investigate the nanoscale defect density variations in the cutting edge area. At the cutting edge, interface-like planar domains of high scattered intensity were found, while a gradual increase of the SAXS intensity at the rake face was correlated with pores found by scanning electron microscopy. Furthermore, the coating’s é111ê fibre texture orientation correl- ates with the surface normal of the substrate, with abrupt orientation changes across the aforementioned interface-like structures. The planar regions next to the edge exhibit gradual and constant stress profiles on the flank and rake faces, respectively, with anisotropic defect build-ups. Directly at the edge, nonlinear lateral and cross-sectional compressive residual stress gradients ranging from∼0to−3GPa were observed, which, together with the interface-like planar domains, may represent a reliability issue during operation.

2. A bilayer Cr-CrN thin film was subjected to scratch testing to simulate a single-asperity contact and evaluate the stresses induced by elasto-plastic de-

(22)

formation accompanying the contact. The experimental results presented in Paper B revealed a complex alternation of the columnar grain microstructure and the formation of pronounced lateral and through-thickness stress gradients, which are interpreted by a ﬁnite element model. After scratching, the in-plane stresses within the CrN toplayer relax from ∼ −3.5 to ∼ 0GPa at the surface along the scratch track, while they increase up to −6GPa at a distance of about 6µm away from the imprint centre. Within the ductile Cr interlayer, compressive and tensile stresses of up to−1.5 and 1.5 GPa formed at the imprint centre and in the surrounding pile-ups, respectively. Also, the evaluated out-of-plane and shear stress distributions exhibit steep gradients, which are correlated with the nanoscopic microstructural changes observed by transmission electron microscopy, transmission Kikuchi diffraction, SAXSM and peak width analysis. Within the brittle CrN, the scratch test results in grain sliding and the formation of nanoscopic intragranular defects. The Cr interlayer’s thickness reduction and pile-up formation are accompanied by a bending of columnar crystallites and localized plastic deformation. In summary, the quantitative stress data elucidate the stabilizing role of the ductile Cr sublayer, which suppresses catastrophic cross-sectional fracture during scratch tests.

3. In order to understand better the nanoscale fracture response of thin films and their effect on functional properties, new high-resolution in situ investigation tools need to be developed. In Paper C, a clamped cantilever with dimensions of 200µm×23.6µm×40µm was cut by focused ion beam milling from a thin film composed of, in total four, alternating CrN and Cr sublayers on high strength steel. The clamped cantilever was loaded in two steps to 460 mN and four two-dimensional strain maps were obtained by in situ cross- sectional X-ray nanodiffraction. Preliminary ex situ characterisation revealed the depth variation of fibre texture within CrN and Cr, respectively, allowing also the depth-dependent alteration of stiffness to be evaluated, as well as residual stress gradients within the individual layers. The in situ diffraction experiment revealed (i) a strong influence of the residual stresses on the deformation behaviour, (ii) the crack arrest capability of the CrN-Cr interface, and (iii) a crack tip blunting effect. In detail, effective stress intensity of

−5.9±0.4MPa m¹^/² arose in the notched Cr layer, as a consequence of the pre- existing residual stress state. Crack growth within the Cr sublayer occurred at a critical stress intensity of2.8±0.5MPa m¹^/². At 460 mN, after crack growth

(23)

to the adjacent CrN-Cr interface, the stress ﬁelds of the crack vanished. The results were furthermore complemented by two-dimensional numerical simulation using the eigenstrain reconstruction method, to gain further insight into the stress evolution during loading. Finally, the results represent an important step towards the understanding of the fracture behaviour of thin ﬁlms composed of alternating brittle and ductile materials.

4. In order to address the aspect of thermal stability and of temperature-dependent changes in hard protective thin ﬁlms, the decomposition routes of metastable c-AlCrN were assessed in the temperature range of 25 to 1100°C, using the newly developed multi-parameterin situhigh-energy high-temperature grazing incidence transmission X-ray diffraction (HE-HT-GIT-XRD) method. The onset temperature of decomposition of c-AlCrN into w-Al(Cr)N and c-Cr(Al)N was indirectly proportional to the as-deposited residual stress magnitude, whereas the onset stress of decomposition was found to be independent of the thermo-mechanical history of the investigated thin films. The methodology and results were published in Paper D. The newly developed HE-HT- GIT-XRD method was also employed to investigate the thermal stability of microstructure and phases of a biomimetic self-assembled hard and tough CVD TiAlN thin film investigated in Paper E.

5. In Paper E, the mechanical and thermal properties of a biomimetic self- assembled hard and tough TiAlN thin film consisting of six hierarchical levels have been investigated extensively. The 2.7µm thick film was formed by chemical vapour deposition by a variation of two different gaseous precursors and through bottom-up self-assembly in only ∼15mins of deposition time, which resulted in an irregularly arranged hard and tough multilayer stack. The hard sublayers consisted of herringbone-shaped micrograins, while tough interlayers were composed of spherical nanograins, forming a stack of lamellar nanostruc- tures of alternating coherent/incoherent, hard/tough, single-/poly-crystalline platelets. Thermal stability was studied by the HT-HE-GIT-XRD method presented in Paper D and revealed microstructural and phase stability up to 900 and 950°C, respectively. Intrinsic toughening mechanisms mediated by five different types of interfaces resulting in inter- and transgranular fracture modes with zigzag-like crack patterns at multiple length-scales were revealed by micro- and nanomechanical testing, performed in situ in scanning and transmission electron microscopes. Hardness, fracture stress and toughness of 31 GPa, 7.9 GPa and 4.7 MPa m¹^/² were evaluated. The film’s thermal, mi-

(24)

crostructural and mechanical characteristics represent a breakthrough in the production of hard protective, wear-resistant thin ﬁlms.

In the following chapters, the characteristics of the investigated thin ﬁlms and the analytical techniques employed in the appended publications are introduced and brieﬂy discussed.

(25)

2

Hard protective thin films

The demands on properties of hard protective thin films are set by the metal cutting industry and films therefore have to be continuously improved to meet their growing requirements. High hardness, together with sufficient toughness, is necessary for wear resistant thin films [61]. Hard protective thin films composed of transition metal nitrides are characterized by their mixed bonding state (cf. Fig. 2.1), which is tuneable by alloying between metallic, covalent and ionic bonding [62].

Figure 2.1.: Schematic sketch of possible bonding types and properties associated with them, along with the approximate placement of selected nitride ceramics within this context. Reproduced from Ref. [62].

In this section, the main characteristics of the hard protective nitride thin ﬁlms investigated in this thesis are outlined.

(26)

2.1. TiN

TiN has been introduced nearly half a century ago as a material for hard protective thin ﬁlms and is still relevant in thin ﬁlm research and industry. Nowadays it is mostly used as adhesion layer, diffusion barrier and as marker for wear, the latter due to its bright golden colour [63]. The oxidation resistance of TiN is limited, and oxidation starts already at ∼550°C, when the poorly-adherent porous rutile TiO2

phase is formed [64–67]. Additionally, mechanical properties are quite moderate, as reported values for hardness lie in the range of ∼ 18-25 GPa [2,61] and those for fracture toughness in the range of 2-3MPa m¹^/² [44, 68].

However, for the analysis of deposition-process-induced nanoscale stress and microstructure gradients, it is perfectly suitable, due to the facts, that it crystallizes in the B1 NaCl structure over a relatively wide Ti/N ratio [69] and its elastic properties are quite isotropic [70]. Additionally, during deposition using CAE moderate compressive stresses are introduced in the coating, which is in contrast to Al-alloyed thin films (see for example the as-deposited stress magnitudes of c-AlCrN (Paper D) and TiN (Paper A) of −4.3 and −1.7GPa, respectively, on flat surfaces when applying the same deposition conditions ).

2.2. Cr/CrN multilayer thin films

Significant scientific interest was drawn during recent years on CrN-based thin films.

The mechanical and structural properties of CrN are quite similar to the ones observed for TiN, which is a hardness of 18-29 GPa and that it crystallizes in the cubic B1 NaCl structure [71–74]. During annealing in vacuum, N diffuses out of the lattice and in a two-step process hexagonalCr₂Nand finally metallic Cr are formed at temperatures above∼930°C and 1100°C, respectively [75]. However, the oxidation resistance of CrN is higher compared to TiN, which can be related to the formation of dense and passivatingCr₂O₃ instead of porous rutile [9, 65]. In contrast to TiN, the elastic modulus of CrN is highly anisotropic, ranging from∼230GPa along the é111ê orientation to ∼ 230GPa along the é100ê orientation [70]. The compressive residual stress magnitude, the grain size, the texture and the thermal expansion coefficient of MSD CrN films can be varied in a broad range, depending on the applied deposition conditions [76, 77].

The formation of a é100êfibre texture and accompanying increase of stiffness can be triggered by the incorporation of ductile Cr interlayers and results in a pseudo- epitaxial relationship, which can be expressed as {100}CrNë{100}Cr [78]. Using

(27)

this, a bilayer film, consisting of 1µm CrN and 2µm Cr on high-speed steel was designed, for which a wear rate reduced by an order of magnitude compared to a 3µm thick CrN thin film was revealed, despite the fact, that the thickness of the wear resistant nanoceramic CrN layer was reduced accordingly [79]. Furthermore, alternating Cr and CrN have been proven (i) to prevent catastrophic failure of the film during indentation, by dissipating cracks at the Cr/CrN interfaces [80] and (ii) to simultaneously increase the stiffness, fracture stress and fracture toughness of Cr/CrN multilayer thin films tremendously [43, 44].

Clarifying the influence of interfaces separating sublayers of distinctly different properties (brittle/hard nanoceramic CrN and ductile/soft nanocrystalline Cr) on the scratch and fracture resistance of a hard protective thin film, and resolving the accompanying nanoscale microstructure and stress gradients were the main focus of Paper B and Paper C.

2.3. AlCrN

The replacement of Cr with Al in Al_xCr_1−xN over a wide compositional range is generally beneficial, since it leads to the formation of a metastable solid solution with increased hardness, oxidation and wear resistance [81, 82]. However, when the solubility limit (x . 0.7) for Al in the cubic (c) B1 lattice is exceeded or when the metastable solid solution is heated above 800-900°C, wurtzite (w) B4 AlCrN is formed, which is generally detrimental to the mechanical properties and oxidation resistance [83–85]. Upon annealing in vacuum, the decomposition path of metastable c-AlCrN thin films is well-known [86–89] and can be expressed as follows

c−AlCrN−−−−−−−→⁷⁰⁰⁻^900°C c−Cr(Al)N + w−Al(Cr)N

900−1000°C

−−−−−−−→c−Cr(Al)N + w−Al(CrN) 1000−1200°C

−−−−−−−−→c−Cr(Al)N +w−Al(Cr)N + h−Cr₂N + N₂ 1000−1200°C

−−−−−−−−→+w−Al(Cr)N + Cr + N₂. However, these data were obtained from AlCrN powders by a combination of in situ differential scanning calorimetry and thermo-gravimetral analysis in vacuum, combined with ex situ microstructure and phase analysis by transmission electron microscopy and laboratory XRD [86–89], respectively. Therefore, they are inherently not able to explore the influence of microstructure and residual stress on the decomposition process, since residual stress is released and diffusion paths are shortened in powdered thin films. Exceptionally, powders of other thin film materials, such as

(28)

TiAlN [90], ZrAlN [91] and TiCrAlN [92], as well as thin lamellas of TiAlN [93] and TiZrAlN [94] have been investigated by means ofin situ synchrotron XRD, to study several temperature-dependent phenomena, such as evolution of phases, lattice parameters and/or in-plane strains. However, these experiments were mainly focussed on the investigation of complex decomposition routes and the related lattice parameter changes. In total, the complex interplay between microstructure/residual strains originating from the deposition process and the temperature-dependent multi-stage decomposition process of transition metal nitrides is still unexplored, which was the main motivation behind Paper D.

2.4. AlTiN

Al_xTi_1−xN was proposed as an alternative to TiN approximately 35 years ago [65].

The suggestion is that Ti is replaced in the cubic B1 NaCl lattice by Al, forming a metastable solid solution with a solubility limit of x . 0.67 [69] or . 0.8-0.9 [95] when synthesized using PVD or CVD, respectively. Generally, also for TiN, the replacement of Ti with Al leads to a pronounced increase of hardness, up to

∼ 35GPa [96], while exceeding the solubility limit promotes the formation of w- Al(Ti)N, which is detrimental to mechanical properties [97–101]. To increase the oxidation resistance, however, high Al contents are necessary and prevent the formation of porous and poorly adherentTiO₂, but again, formation of w-Al(Ti)N is also considered unfavourable [102, 103]. Furthermore, during annealing to temperatures above ∼850-900°C, the metastable cubic solid solution decomposes via a spinodal decomposition into c-TiN and c-AlN, which leads to a nanoscale structuring and a further increase of hardness [96, 98] and also fracture toughness [104]. However, when annealing to even higher temperatures, w-Al(Ti)N is formed, which again leads to a deterioration of mechanical properties [96, 98, 104].

Here, the further attention was focussed on recent research progress concern- ing AlxTi1−xN synthesized by low-pressure CVD [3, 105–107], which forms self- organized nanolamellar structures of different shapes and sizes depending on the Al- content. When synthesized with an Al-content of x = 0.95, the particular growth conditions resulted in randomly oriented cubes of the self-organized nanolamellar structure with a size of ∼ 100nm embedded in a c-TiAlN matrix [106], where the lamellae consisted of alternating incoherent 2 and 11 nm thick layers of w-Al(Ti)N and c-Ti(Al)N, respectively. The film formed with such a microstructure exhibited a moderate hardness of 27 GPa, but superior oxidation resistance up to 1100°C.

By slightly reducing the Al-content to x = 0.8, a different film is grown, consist-

(29)

ing of irregularly formed stacks of herringbone crystallites extending over the whole coating thickness without interruption of nanocrystalline matrix regions [3, 107].

The herringbone crystallites are composed of coherent lamellae consisting of 1.5 and 11 nm thick c-TiN and c-AlN layers, respectively, resulting in outstanding mechanical properties with a peak hardness of 37 GPa and slightly reduced oxidation resistance. These two unique microstructures, differing significantly in mechanical properties, can then be formed alternatingly, by a simple alternation of the pre- cursor compositions in the CVD process [106, 107]. The multi-layered hierarchically structured thin film system produced thus opens up huge potential for the advance- ment of hard protective thin films, which was the main encouragement for the work carried out in Paper E.

(30)

(31)

3

Selected advanced characterization techniques

3.1. X-ray diffraction

X-ray diffraction is a versatile and capable tool for the structural characterization of materials and therefore widely used in materials science [108–112]. It can be used to identify individual crystal structures within the investigated volume and thus also (i) thephase composition, (ii) the phase-specific crystallite orientation distribution (so-called texture), (iii) the orientation-dependent shape and size of coherently diffracting domains, defect density and strains of 2^nd and 3^rd order, qualitatively assessed by the full width at half maximum (FWHM) of diffraction peaks and (iv) to perform a quantitative analysis ofresidual strainsof1^storder. Generally, XRD is limited to crystalline materials, which however includes most ceramic and metallic thin films.

Interference of coherent waves scattered by a periodic structure, e.g. a crystal, is called diffraction [108–112]. Independent of the X-ray source and measurement geometry, the basic principle for investigations using XRD is Bragg’s law. Fig. 3.1 presents a geometric representation of Bragg’s law, where the elastic scattering of two parallel waves of an X-ray beam on two parallel planes of a crystal lattice and subsequent constructive interference of the scattered waves at an angle 2θ with respect to the incident wave vector can be observed. Bragg’s law is defined as follows

2dsinθ=nλ, (3.1)

wheredis the lattice parameter,θis Bragg’s angle,λis the X-ray wavelength and n is an integer giving the order of the diffraction peak. Constructive interference of the two scattered waves, i.e. diffraction, is observed, when the phase shift between the waves is a multiple of the X-ray wavelengthλ.

A set of observed peaks corresponding to a set of Bragg angles (and hence also

(32)

Figure 3.1.: Schematic representation of the Bragg condition for a primitive square lattice.

The incident X-rays described by a wave vectorK0 are scattered by atoms, which results in a phase shift of2dsinθfor waves scattered at adjacent planes.

Maximal intensity is only observed, when the phase shift is a multiple of the wavelengthλ, resulting in a diffracted wave with wave vectorK1. Reproduced from [111, 112].

lattice parameters) is usually distinct for every crystalline phase withine.g. a polycrystalline multi-phase material [108–112]. However, there are materials with similar crystal structures, which differ only slightly in their lattice constants, e.g. Fe and Cr [113] or Al2O3 and Cr2O3 [114].

Given that the material is (i) polycrystalline with individual crystallites oriented randomly and that (ii) the incident X-ray beam is monochromatic, the sequence of lattice planes results in a sequence of (powder) diffraction cones.

These diffraction cones are often assessed by a symmetric (θ−2θ) scan using a laboratory XRD instrument, resulting in a plot of diffraction intensityI vs. diffraction angle2θ. However, when employing this reflection geometry [111, 115], several restrictions are imposed by it. First and foremost, it is very challenging and usually not unambiguously possible, to deconvolute signal contributions from different pen- etration depths of the X-ray (dependent on the incidence and exit angles), which is an intrinsic problem of the reflection geometry. Second, the brilliance, a product of several parameters describing the radiation source, including photon flux density, monochromaticity and beam parallelity, of laboratory X-ray sources is limited. A high brilliance of the X-ray source is however necessary in order to (i) provide beam diameters below the micrometre-regime at a sufficient flux, since beam intensity and focussing are mutually exclusive parameters, due to low efficiency of X-ray optics

(33)

[1, 115–117] and to (ii) provide short exposure times for in situ experiments, for instance when high temperature (time) resolution is necessary, during annealing of thin ﬁlms [118].

Hence, to perform XRD measurements with small X-ray beams, it is necessary to use synchrotron light sources producing X-ray beams with brilliances orders of magnitude higher compared to the ones available in the laboratory. Furthermore, with such high brilliance, it is also possible to employ transmission geometries for the measurements, which is a key feature for the technique of cross-sectional X-ray nanodiffraction presented hereafter.

3.1.1. Cross-sectional X-ray nanodiffraction

In order to assess nanoscale gradients in thin films originating from the deposition process or through externally applied loads, high resolution characterization tools are necessary. Since 2012, cross-sectional X-ray nanodiffraction (CSnanoXRD) has been developed and advanced, resulting in a minimum achievable size of the incident X-ray beam of even below 30 nm [1, 116]. In prior studies, CSnanoXRD has been proven to be successful at revealing gradients of phases, crystallite size and shape, texture and strains in thin films in as-deposited state [4, 117, 119], after annealing [120], after oxidation [121] or after ex situ indentation [80]. Furthermore, CSnan- oXRD using a spatial resolution down to 200 nm coupled with anin situindentation device has been developed and successfully utilized to probe the stress fields emanating from indentation of TiN and CrN thin film featuring sublayers of alternated deposition conditions resulting in monophasic/microstructural interfaces [57, 122].

Given the available literature, CSnanoXRD shows great potential in resolving the problems regarding above mentioned nanoscale gradients and give novel insights into the structure-property relationships of thin films. In the following section, the methodology of CSnanoXRD, which was used in Papers A, B, C and E, is briefly discussed.

3.1.1.1 Phase analysis

One of the basic tasks of the analysis of structure-property relationships in thin films is the characterization of phases present in the thin film. Since the set of diffraction angles is distinct for every phase, the phase analysis can be performed by a comparison of the experimentally assessedI−2θplot with a model diffractogram composed of diffraction lines from the Powder Diffraction Files (PDF) database from the International Centre for Diffraction Data (ICDD, formerly JCPDS) [123, 124].

(34)

The differences between the experimentally assessed and the modeled diffractogram give then important information about the microstructure and morphology of the thin films, which will be discussed later.

Figure 3.2.: Schematics of a typical CSnanoXRD experiment. A monochromatic X-ray beam, for instance focused using multi-layer Laue lenses [1, 125], with a dia- meter in the range 30-50 nm irradiates the sample, which is placed in the beam’s focal spot (i.e. where the beam dimensions are minimal). The 2D detector, a charged coupled device (CCD), intersects the diffraction cones emanating from the sample resulting in a diffraction pattern composed of Debye- Scherrer rings. The sample is scanned perpendicular to the beam resulting in either one- (z-scan) or two-dimensional (y-z-scan) maps of diffraction patterns recorded and subsequently evaluated. (own unpublished work)

A typical CSnanoXRD experiment is performed either as one-dimensional (i.e.

cross-sectional) scan along the z-axis or as a two-dimensional (2D) mesh scan by moving the sample along they- andz-axis with a defined step size(s). The 2D detector intersects the diffraction cones (cf. Fig. 3.2), which results in a 2D diffraction pattern composed of Debye-Scherrer rings emanating from the individual phases within the thin film at every measurement point during the experiment. When a

(35)

4µm thick thin ﬁlm is investigated with a step size of 50 nm in a cross-sectional manner, 80 diffractograms are recorded. The collected data needs to be integrated azimuthally in order to perform phase analysis. Dedicated programs, which are widely used in the synchrotron X-ray community, are Fit2D [126] and PyFAI [127, 128].

A prerequisite for correct azimuthal integration is determination of the precise position and orientation of the 2D detector with respect to the beam and the sample.

The (i) sample-to-detector distance, (ii) the beam centre on the detector, (iii) the rotational misalignment and (iv) the detector tilt are obtained by measuring a samples, which yields near-ideal and well-known diffraction patterns, a so-called calibration standard. Calibration standards are, for instance,Al₂O₃ orLaB₆ powders prepared by the National Institute for Standards and Technology (NIST). The software used for azimuthal integration provides procedures that extract these detector geometry parameters from such calibration measurements and subsequently accounts for them during further processing.

Figure 3.3.: Exemplary phase plot obtained from CSnanoXRD performed on an∼3.2µm thick AlCrN film deposited by CAE on a cemented carbide substrate and annealed in vacuum at 1000°C for 1 h. 2θ is the diffraction angle and the diffraction intensities are colour-coded. Peaks of specific phases are marked with Miller indices of the corresponding lattice plane families. (own unpublished work)

After azimuthal integration, I−2θ datasets are obtained for every measurement point of the scan, are collated and finally plotted against the scanned coordinate,

(36)

which results in phase plots, where the diffracted intensities are colour-coded (cf.

Fig. 3.3). The individual diffraction peaks obtained at different scan positions can then be related to the phases present in the film at the respective film thickness. In the example given in Fig. 3.3, an AlCrN film annealed in vacuum at 1000°C for 1 h is presented, along with the tabulated reflections for c-CrN, w-AlN and hexagonal (h)Cr₂Nfrom the PDF database [129]. During an annealing experiment at 1000°C, the metastable c-CrAlN phase decomposes into c-CrN, w-AlN and h-Cr2N and N2, which is reflected by the increasing intensity of w-AlN and h-Cr2N peaks towards the film surface. Furthermore, it can be seen, that the w-AlN and c-CrN reflections are not exactly at their indexed positions, which indicates, that the wurtzite and cubic phases are not stoichiometric.

3.1.1.2 Small-angle X-ray scattering microscopy

Unlike diffraction, which originates from the lattice periodicity of a material, dif- fuse scattering at small diffraction angles (θ ∼0.1-10deg, depending on the X-ray wavelength) is sensitive to significantly longer-range variations of electron density within the X-ray gauge volume, such as nanoscale alternation of phases, grain boundaries, interfaces, cracks, pores, etc. The observable nanoscopic structural heterogeneities are of the size of ∼^λ/θ [130]. Especially in deformed sample regions or in nonuniform sample environments, changes of this small-angle signal are to be expected, as they are correlated with structural and mechanical gradients. A qualitative value is obtained for every point within a measurement area, by azimuthal (δ) and radial (θ) integration of the scattered intensity I(θ, δ), as follows

I =

Ú Ú _δ=360^◦_,θ₁

δ=0^◦,θ0

I(θ, δ)dθdδ (3.2)

The resulting integrated small-angle X-ray scattering (SAXS) intensity is then a measure for the amount of structural heterogeneities within the X-ray gauge volume.

By scanning/mapping a sample region, SAXS microscopy (SAXSM) can thus be performed.

An exemplary scan and map are presented in Figs. 3.4 and 3.5, respectively. In Fig. 3.4, the integrated SAXS intensity obtained from the same AlCrN coating as in Fig. 3.3 is presented. The highest scattered intensity is found close to the coating- substrate interface, with a strong decay with increasing film thickness. This can be related to the small grain size and high interface density within the nucleation zone close to the substrate. In Fig. 3.5, the SAXS micrograph obtained from the cutting edge area of a TiN coating on a cemented carbide substrate is presented [131]. Here,

(37)

Figure 3.4.: Integrated SAXS signal obtained from the∼3.2µm thick AlCrN ﬁlm presented in Fig. 3.3. A local rise of scattered intensity can be seen at depths of 0.5 and 3.3µm corresponding approximately to the surface and the substrate coating interface, respectively. A minimum of scattered signal was found at a depth of∼1.25µm. The increase in scattered intensity can be correlated to the higher diffracted intensity ofCr2N, which corresponds to a higher interface density (smaller crystallites). (own unpublished work)

a gradually increasing SAXS intensity towards the rake face can be correlated to the promoted formation of pores at the rake face and the lines of increased intensity directly at the edge are related to planar domains in the coating, separated by interface-like under-dense structures [131]. Especially in the latter case, SAXSM has been proven to be useful to obtain data about the coating’s microstructure, which are furthermore directly correlated to gradients of texture and residual stress [131].

(38)

Figure 3.5.: Integrated SAXS signal obtained from a TiN hard coating deposited onto cemented carbide cutting edge. The substrate-coating interface and the surface can be clearly detected due to their comparatively higher scattering intensity.

Additionally, a gradual intensity increase from the ﬂank to the rake face was detected, corresponding to an increase of structural defects along the rake face. Furthermore, the lines of increased scattered intensity, visible within the edge region, could be related to planar domains within the coating, separated by an interface like morphological feature. Reproduced from [131].

3.1.1.3 Texture analysis

Only in rare cases, the distribution of the crystallite orientations in a polycrystalline thin ﬁlm is random. Texture of materials can span over the whole range of ordering, from solids where crystallites have nearly the same alignment (high mosaicity, sharp texture) to materials where it is nearly impossible to distinguish the diffraction pattern of the solid from that of a powder [111, 112].

Furthermore, many important mechanical and functional properties of materials are also anisotropic, i.e. dependent on crystal orientation. A sound example is the anisotropic Young’s modulus of CrN, which varies between∼230and 450 GPa,

(39)

when considering the é111ê oré100êorientations [70], respectively.

In laboratory XRD, crystallographic texture is investigated by considering the variation of diffraction intensities with respect to the orientation of the investigated sample. During the measurement, the diffraction angle 2θ of a certain reflection is fixed, while the sample is rotated around two axes [108, 109, 111, 112]. The recorded intensity distribution of the investigated reflection is then proportional to the orientation frequency of the chosen lattice plane family with respect to the sample coordinate system. The common representation for this orientation distribution is the pole figure, which is a stereographic projection of the lattice plane normal vec- tors onto a plane. The orientation-dependent diffraction intensities are then plotted in a colour-coded fashion.

It can be seen in Fig. 3.6 that during a CSnanoXRD experiment performed in transmission geometry the orientation information contained in a Debye-Scherrer ring is rather limited. The intensity distribution along the ring represents a cut

Figure 3.6.: Schematic representation of the orientation relationship between the out-of- plane centred pole figure and the diffraction intensity distribution I(θ, δ) recored along a Debye-Scherrer ring on a 2D detector in transmission geometry.

Pole figure and detector image are taken from an experiment performed on a TiN coating deposited on a cemented carbide cutting edge reproduced form [131].

(40)

through the corresponding pole ﬁgure (cf. Fig. 3.6). Without further assumptions, the texture of a thin ﬁlm material thus cannot be evaluated unambiguously.

However, with regards to thin films, two types of texture are dominant, the so- called biaxial texture and the fibre texture, where the latter is observed in most hard protective thin films. When a thin film develops a fibre texture, all crystallites have a fixed orientation in the direction of the fibre axis and one remaining orientational degree of freedom, allowing for a rotation around the fibre axis. In nearly all cases of thin films studied, the fibre axis is parallel to the surface normal, except, when the deposition flux is strongly inclined, as in glancing-angle deposition (cf. [2, 48, 49]).

Since most thin films develop the above-mentioned fibre texture during growth, the pole figure can be reconstructed for the thin film material of interest, supposing rotational symmetry around the out-of-plane sample axis. The reconstruction is performed by considering the geometrical relationship between the coordinates of a point on the 2D detector and the coordinates of the same point on the corresponding out-of-plane pole figure (cf. Fig. 3.6). This geometrical relationship is expressed as follows

sinα= sinδcosθ (3.3)

and

cosβ= +

ó cos²δcos²θ

1−sin²δcos²θ | −90°< δ <90°, (3.4)

−

ó cos²δcos²θ

1−sin²δcos²θ |90^◦ < δ <270°,

whereδ and 2θare the azimuthal and radial (diffraction) angle on the 2D detector, respectively, while α and β represent the radial and azimuthal coordinates of the pole figure, respectively (cf. Fig. 3.6) [132]. An example for an out-of-plane pole figure and the corresponding reconstruction from a synchrotron detector image of a TiN film deposited on a WC-Co cutting edge can be seen in Fig. 3.7a and 3.7b, respectively.

Having confirmed the fibre nature of the thin films texture, the preferred orientation of the thin film along the cutting edge can be evaluated by simply considering the orientation δ_p¹¹¹ of the maximum intensity of the Debye-Scherrer ring for every dataset obtained during the experiment (cf. Fig. 3.8).

(41)

Figure 3.7.: Pole figures evaluated from the 111 reflection of a TiN thin film. A pole figure obtained by a laboratory measurement (a) and a pole figure reconstructed from 2D data obtained during the corresponding synchrotron measurement in transmission geometry (b) are presented. Reproduced from [131].

Figure 3.8.: 2D contour plot of the orientationδp¹¹¹of theé111êﬁbre texture. Reproduced from [131].

(42)

3.1.1.4 FWHM analysis

When an ideally parallel and monochromatic X-ray beam interacts with a single coherently diffracting domain of a hypothetical ideal material having infinite di- mension, which is furthermore free of any kind of lattice defects, a perfectly sharp diffraction peak with negligible width would emerge. Any kind of deviation from these idealisations will lead to an increase of the peak width. The measure to de- scribe the peak width is the full width at half maximum (FWHM)β. Contributions to the FWHM are the size of coherently diffracting domains, gradients of strains of 1^st order (macrostrains) and strains of 2^nd and 3^rd order, which are, in general, lattice defects, such as dislocations and lattice distortions.

In order to quantify the FWHM of Debye-Scherrer rings, it is necessary to fit the diffraction peaks to analytical functions. Therefore, the 2D diffraction patterns collected during the CSnanoXRD experiment are integrated azimuthally, similar to the procedure described for phase analysis (cf. Sec. 3.1.1.1). However, the main difference is that the integration is performed inn circular sections of a predefined azimuthal widthw= ³⁶⁰_n deg, so-called cakes. The result is a set ofI(2θ, δi)proﬁles, whereδ_i =iw is the azimuthal orientation of the centre of thei-th cake and which can be used for orientation-dependent peak ﬁtting.

Typical functions which are used for peak fitting are the Lorentzian or Gaussian functions. Generally, the use of the pseudo-Voigt function, which is a linear combination of the former two, is beneficial. The pseudo-Voigt function, modified to diffraction-related quantities, reads as follows, for every azimuthal section,

I(2θ, δ_i) =µ I₀ 1 +

32θ−2θ0(δi)

1 2β(δi)

42 + (1−µ)I₀e 5

−ln 2

12θ−2θ0(δi) 12β(δi)

226

, (3.5)

where I(2θ, δ_i) is the observed diffraction angle-dependent intensity at the azimuthal orientation δ_i, 2θ₀(δ_i) is the peak’s centre position, I₀ is the maximum intensity of the peak at 2θ0 , β(δi) is the orientation dependent FWHM of the peak and µis the ratio of the Lorentzian over Gaussian distributions contributing to the peak function.

The Scherrer equation [108, 109, 111, 112] describes the inverse relationship between the width of a diffraction peak (or the corresponding Debye-Scherrer ring) and the average size of coherently diffracting domains, which is roughly similar to the grain size of a polycrystalline material. The Scherrer equation is expressed as follows,